Predicting Patient Costs

Healthcare spending in the US. reaching an all-time high, but new payment methods have the potential to alleviate some of the cost. One example is prospective payment, in which health care providers receive a dollar amount for services ahead of time to incentive efficient and effective delivery of care. The effective implementation of such a method would rely on models that effectively predict patient costs.

My partners and I teamed with the University of Chicago's Chief of Hospital Medicine to develop these patient cost-prediction models. Using electronic health records from over 34,000 patients from University of Chicago Hospitals we extracted cost-predictive variables, trained and evaluated predictive models, and identified variables highly predictive of future cost.

View our presentation to the Chicago Chapter of the American Statistical Association, my poster for Joint Statistical Meeting 2016, or my Statistical Significance piece describing the value of statistics in healthcare.